With the growth in smartphone use and increasing demands from patients for immediate access to web-based services, mobile health (mHealth) apps that allow patients to actively manage their own health through mobile and wireless technologies are on the rise [1-3]. Internationally, the popularity of mHealth apps to support the achievement of health objectives is increasing [4], especially for chronically ill patients [1]. For this group of patients, research suggests that mHealth apps lead to increased confidence in disease management [5,6], improved therapy compliance, better health care outcomes [7], and even reduced costs [6].

Despite the proven impact of mHealth apps on patient well-being [8], patients do not always show high levels of behavioral engagement with them [9-11]. Here, behavioral engagement refers to the adoption and continued usage of mHealth apps by chronically ill adults. User data from popular app stores even show that most mHealth apps are only used a few times before being abandoned [12]. Less than a third of chronically ill patients aged 50 years and older currently use an mHealth app. More than a third of patients have used mHealth apps in the past but have then stopped using them [13]. In an attempt to better understand why behavioral engagement with mHealth apps is low, we explore the drivers of behavioral engagement with mHealth apps.

In line with the technology acceptance model and the unified theory of acceptance and use of technology, several researchers have, in recent years, pointed out the importance of social-related and technology-related factors when explaining behavioral engagement with health care technologies [14,15]. In light of the behavioral engagement with mHealth apps, there has been considerable research on effort and performance expectancy [16]. However, less research has been dedicated to the role of social and app-related drivers of behavioral engagement with mHealth apps. Pham et al [17] also call for more research on the relationship between mHealth engagement and well-being for chronically ill patients.

Against this background, this research characterizes social enablers (ie, physician recommendation) and app-related enablers (ie, app integration) of behavioral engagement with mHealth apps among chronically ill patients, thereby also considering the impact of behavioral engagement with mHealth apps for patient well-being (that is, hedonic well-being defined as the extent of pleasure and eudaemonic well-being defined as the extent of self-efficacy [18]).

In this study, we rely upon a scenario-based survey study, which is very common in business research [42] and in technology acceptance studies [43]; an increasing use of scenario-based surveys are also being used in health care [44]. Scenario-based survey studies have the advantage of eliminating the difficulties associated with observation or enactment of events in real life, such as in this study, with undesirable outcomes, and with not reaching a sufficiently large sample size, as can happen when forcing patients to use a nonintegrated app [42]. Compared to recall-based surveys, scenario-based surveys also have the advantage of reducing biases from memory retrieval [45].

This scenario-based survey study involves a between-participant 2×2 design and introduces participants to a scenario. In all scenarios, the patient receives a pamphlet with information on a fictional mHealth app, but the scenarios differed in terms of the recommendation by the physician to use the mHealth app (ie, strong vs weak recommendation to use the mHealth app) and app integration (integrated vs nonintegrated mHealth app). This 2×2 design has 4 different possible scenarios, and each participant was randomly assigned to 1 of these 4 scenarios. After reading the scenario, the participant filled out a questionnaire. The scenarios are detailed in Multimedia Appendix 1.

G*Power 3.1.9 (Heinrich Heine Universität) was used to calculate the required sample size for detecting a medium effect (Cohen d=0.5) in an independent sample t-test (2-tailed). With 80% power at an α level of .05, a total sample size of 204 participants (51 per group) was needed to test the hypotheses. To achieve the required sample size, respondents with chronic conditions were recruited by more than 60 organizations representing the interests of chronically ill people in the Flemish region of Belgium through sharing the survey in their e-newsletter, website, or Facebook page. Eligible respondents (1) had been diagnosed by their physician with a chronic disease and (2) were aged between 18 and 65 years. These age boundaries were set because the empirical literature identified strong differences in the adoption of technology among young people, adults, and older adults [46]. In total, 722 respondents completed the questionnaire. After quality checks (including age and condition checks and a control question), 521 respondents were retained.

The Ghent University Hospital review board approved the study protocol (2019/1975-670202042704), and participants were asked for consent.

We conducted a web-based survey from March to May 2019. Study data were collected and managed using REDCap electronic data capture tools hosted at Ghent University Hospital [47,48]. The survey involved 5 different constructs, including skip patterns. All constructs were measured using previously validated multi-item scales with proven validity and reliability (See Multimedia Appendix 2 [49-53]). The original scales were translated into Dutch using the forward and backward translation technique. Although validated by previous research, the measurement instrument was further tested to ensure reliability within the study context. Cronbach α values of the validated constructs ranged from .748 to .952 and showed that the reliability requirements were met. Reponses were provided using a 7-point Likert scale, with anchors ranging from 1 (“strongly disagree”) to 7 (“strongly agree”). Finally, the survey included questions about age, gender, and the duration of the chronic condition, as it is common to include these demographics in research relating to chronic conditions [54].

We assessed the experimental interventions by comparing the mean score on a single item measuring physician recommendation (“my physician recommends me to use this app”) and a single item measuring app integration (“this is an integrated app”) between the different scenarios. The mean differences for both interventions were significant. The mean score on a 7-point Likert scale for physician recommendation was 4.24 in the weak physician recommendation scenario versus 5.47 in the strong physician recommendation scenario (P=.02). The mean score for app integration was 3.67 in the scenario with a nonintegrated app versus 5.87 in the scenario with an integrated app (P=.004).

To simultaneously test all hypotheses (including drivers, consequences, and moderators), we used a mediation approach [55] with Bayesian estimation [56]. As suggested by Iacobucci [57] and Yuan and MacKinnon [56], the following 3 equations were jointly estimated using structural equation modeling in order to test our proposed conceptual model:

in which the BehavioralEngagement_i denotes the individual i’s (i=1 to 521) behavioral engagement with mHealth apps and the WellBeing_di denotes the 2 (d=1 to 2) well-being dimensions: hedonic well-being (d=1) and eudaemonic well-being (d=2). β₁ denotes the effect of the influence of the physician (PhysicianRecommendation_i) on behavioral engagement in order to test H1a, whereas β₂ denotes the effect of app integration (AppIntegration_i) on behavioral engagement in order to test H2a. β_10d represents the effect of behavioral engagement on both hedonic well-being and eudaemonic well-being and respectively enables testing of H3a and H3b. β₃ denotes the moderating effect of patient attachment to traditional care on the impact of physician recommendation on patients’ behavioral engagement (PhysicianRecommendation_iTraditionalCare_i) and allows investigation of H1b. β₅ denotes the moderating effect of patient mobile app experience on the relationship that the app integration has on the patient’s behavioral engagement (AppIntegration_iMobileAppExperience_i) and enables investigation of H2b. The Σ_1i, Σ_2i, and Σ_3di are error terms with intercorrelation ρ which, in line with the well-being literature [58], accounts for the interdependency between hedonic and eudaemonic well-being (see Figure 1). Controls_ci is a vector of control variables, including patient age, gender, and duration of condition.

Because of structural equation modeling, the paths as specified in equations 1 to 2 are modeled in combination with the measurement model. The measurement model provided evidence of construct validity and discriminant validity, and additional tests revealed our data to be free from the common method and collinearity biases (See Multimedia Appendix 2 and Multimedia Appendix 3 [55,56,59-67] for more details). In addition, the model convergence was inspected and revealed evidence of a well-fitting model (see Model Convergence Assessment in Multimedia Appendix 3 for more details). Finally, the structural equation models, in line with the technology adoption literature [49], are linked between the mobile app experience and attachment to traditional care, as shown in Figure 1.

mHealth has several benefits for patients, especially for those who experience chronic conditions [17]. mHealth can enable patients to manage their condition [68], which is crucial for their well-being [69]. However, beginning to use mHealth apps can be challenging, as many patients have neither experience with nor confidence in using them [17,70]. Although health care technology adoption research has typically focused on the need for user friendliness, usefulness, and performance expectancy [16], we complement this line of research by exploring the importance of social factors (here, physician recommendation) and app-related factors (here, app integration) when behaviorally engaging patients with mHealth and when considering how this engagement impacts their well-being.

With regard to the social factors, we found that physician recommendation positively influences behavioral engagement and consequently, patient well-being. The important role laid out for the physician when behaviorally engaging patients with mHealth apps resonates with evidence about their importance in stimulating other types of patient behavior [71] and enhancing patient well-being [72]. Indeed, the physician is in a unique position to motivate their patients to use mHealth apps, especially when they have longstanding relationships with them [73].

Although we have observed that patients who are more inclined to traditional care are less inclined to use mHealth, it is remarkable to note that these patients can also be motivated to use mHealth apps to the same degree as patients who are more open to receiving modern care (with the use of mobile apps). As such, physicians should be careful not to assume that patients who are more accustomed to traditional care will be harder to motivate to use mHealth. This potential bias should not lead to neglect of this part of the patient population, particularly in the (post-)COVID world where virtual care has become a more integral part of traditional care [74]. Furthermore, although we have focused in this research on the motivating role of the physician, other parties (health care organizations, government, etc) can also be of importance in recommending patients to use mHealth apps. Future research could thus focus on marketing strategies that these parties could deploy to encourage mHealth use. Anderson et al [75] have recently called for an increased use of marketing techniques in health care. Indeed, marketing communication can be instrumental in motivating patients to use new technologies and engage in self-care by creating a positive attitude toward health care technology [76,77].

Besides social factors, our results also demonstrate the importance of app integration. Specifically, integrated apps can have an important positive impact on a patients’ behavioral engagement. The major app stores already offer a staggering 350,000 mHealth apps, most of which remain unsuccessful because of limited behavioral engagement [78]. Designing yet another nonintegrated mHealth app can only add to this pile of underutilized apps. By designing mHealth apps, which have meaningful interaction with existing systems used by the patient (eg, appointment scheduler, patient file, health data), behavioral engagement can be enhanced and the continued usage of an mHealth app can be improved.

In line with previous research [22], this study also shows a moderating role for app experience. Patients with more experience in using apps will place greater value on mHealth apps that are integrated with other health platforms. Given that app experience is rising among all parts of the population [79], the importance of offering integrated apps will only increase in the future. Companies who develop mHealth apps and health care organizations who implement them should focus not only on apps’ appearances and capabilities but also show great care in ensuring that apps are integrated into existing health platforms.

Finally, our data were collected in 2019, shortly before the COVID-19 pandemic. The advantages of mHealth apps during a pandemic have been well-documented [74]. Further, numerous papers on the roles of eHealth, telehealth, and telemedicine in delivering health care services to chronically ill patients during the COVID-19 pandemic have been published [80,81]. The COVID-19 pandemic has shown the importance of mHealth as a means of interacting with patients and providing care. It would be interesting to know how the pandemic has changed the way in which patients use and feel about mHealth apps.

Although this study gives clear indications of the importance of physician recommendation and app integration for behavioral engagement and well-being among chronically ill patients, it is not without limitations, and the results should be interpreted accordingly. First, the respondents self-selected to participate in this study. As a result, generalization of the results to a broader population should be done with care. However, because we opted for self-selection, a larger number of participants was recruited. Second, this study uses self-reported data and not actual behavior. This makes the study more vulnerable to self-report biases such as socially desirable answers and information bias. Future research could go beyond scenario-based research and implement the proposed interventions in a randomized controlled trial. Third, this study utilized cross-sectional data, which limits the possibilities of drawing conclusions on causal relationships. Future research could benefit from a longitudinal approach by collecting data at different points in time among the same respondents [12]. It can be envisioned that the engagement of chronically ill patients, in particular, might differ in time, as the severity of their condition or their need for support fluctuates. Fourth, the strength of the relationship between the physician and patient was not included in the study design. Future studies could include measures of this relationship, since the relationship between physician and patient may act as a significant mediator of the relationship between physician recommendation, behavioral engagement with mHealth apps among chronically ill patients, and their well-being.

An ever increasing number of mHealth apps are being developed by both commercial enterprises and health care organizations. Although these apps can have a positive impact on patient well-being, various studies have shown that simply designing an effective app does not guarantee their adoption by users. This study focused on the importance of physician recommendation and app integration in increasing behavioral engagement and well-being among chronically ill patients. It highlights the importance of app developers considering behavioral engagement during and after the development of mHealth apps. During development, attention should be given to ensuring app integration so that communication and interaction with existing health care systems is possible. Integration is an important characteristic that can encourage patients to start using an app—especially when they are experienced app users. After development, it is important to motivate patients to adopt the mHealth app. This study has shown that physicians have an important role to play in motivating chronically ill patients to engage with mHealth apps.